Our goal was to determine whether performance variability during predictive visual tracking can provide a screening measure for mild traumatic brain injury (mTBI). Seventeen subjects with chronic postconcussive syndrome and 9 healthy control subjects were included in this study. Eye movements were recorded with video-oculography as the subject visually tracked a target that moved through a circular trajectory. We compared the variability of gaze positional errors relative to the target with the microstructural integrity of white matter tracts as measured by the fractional anisotropy (FA) parameter of diffusion tensor imaging. Gaze error variability was significantly correlated with the mean FA values of the right anterior corona radiata (ACR) and the left superior cerebellar peduncle, tracts that support spatial processing and sustenance of attention, and the genu of the corpus callosum. Because the ACR and the genu are among the most frequently damaged white matter tracts in mTBI, the correlations imply that gaze error variability during visual tracking may provide a useful screening tool for mTBI. Gaze error variability was also significantly correlated with attention and working memory measures in neurocognitive testing; thus, measurement of visual tracking performance is promising as a fast and practical screening tool for mTBI.